2 ms·
Honestly it’s astounding just how good the representations from top notch virtual instrument libraries can be given the very real limitations of MIDI. The combi
by chriskrycho 5y ago
Honestly it’s astounding just how good the representations from top notch virtual instrument libraries can be given the very real limitations of MIDI. The combination of key-switched articulations with velocity-driven patch variation plus clever legato programming means that a well-practiced DAW programmer can get performances out of libraries that blow my mind. Do they sound like a live performance? Not quite (and less so for solos or small chamber ensembles than for larger ensembles). They sound far better than it would seem they have any right to, though. And the upside to the MIDI encoding is that while you’re absolutely right that it isn’t a natural fit for e.g. violins (where your suggested approach could be a much closer representation), they are universal and therefore much more transferable.
All of that mixed praise of MIDI notwithstanding, I basically agree with the take here. I (self)-published [1] a fanfare I wrote last year taking the path from a score authored in Dorico to MIDI to Logic Pro and the result came out surprisingly well out of the gate, and with much tweaking ended up in something that if I didn’t love, I also didn’t absolutely loathe, which had been my previous experience of these sorts of things.
In any case, the thought of trying to transfer between tools like MuseScore, Dorico, StaffPad, etc. with MIDI just gives me the shudders: it’s absolutely the wrong tool for the job, where (something like) MusicXML is the right tool for the job, because it’s designed from the start to represent notation rather than to represent sound.
[1]: https://songwhip.com/chriskrycho/fanfare-for-a-new-era-of-american-spaceflight https://songwhip.com/chriskrycho/fanfare-for-a-new-era-of-am...
- gnulinux 5y agoOh I think my post is possibly misunderstood? I don't generate wav within MuseScore. I use MuseScore to generate Music XML. Then from that data I have a script that creates relatively more realistic sound using many soundfonts (I created a soundfonts for each of my violin strings, and for each articulation etc). So although the last operation is MIDI->wav I don't use the standard midi rendering. This way I can implement special logic for stuff like triple stops, quadruple stops, expressiveness etc. EDIT: I use midi only to get samples from my sf2 files by using fluidSynth etc.
- chriskrycho 5y agoAh, no, sorry: I took your meaning; I was just talking on a related bunch of things your comment brought to mind! That’s a very interesting approach, though: doing it via script rather than hand-editing in a DAW.